Vernor Vinge - Rainbows End

So he had thought, and so he still hoped. And yet his two young friends' remorseless efforts to help had become the greatest threat to his plans. As today:

"Yes, yes. There are risks," Vaz was saying. "We knew that from the beginning. But letting a serious YGBM project escape detection would be much more dangerous. We must find out what's going on in the San Diego labs. Plan Rabbit can do that."

Keiko Mitsuri shook her head. "Alfred, I have contacts in U.S. intelligence that go back years. These aren't my agents, but they would not tolerate a rogue weapons project. On that, I would trust them with my life. I say we should contact them — very unofficially — and see what they can learn about the San Diego labs."

Alfred leaned forward. "Would you trust them with your country's life? Because that's what we are talking about here. In the worst case, there is not only a YGBM research effort going on in San Diego, but it is supported at the highest levels of the U.S. government. In that case, your friends' best efforts would simply alert their superiors to our suspicions. The evidence would disappear. When it comes to investigating a threat this serious we simply must do it ourselves ."

In one form or another, this was an argument that dated from their Barcelona meeting. Today's installment could be decisive.

Keiko sat back and gave a frustrated shrug. She was presenting in more or less her real appearance and location, a thirty-year-old woman sitting at her desk somewhere in Tokyo. She had transformed one side of Vaz's office with her minimalist furniture and a picture-window view of Tokyo's skyline.

Günberk Braun was less prepossessing. His image simply occupied one of Alfred's office chairs. No doubt Günberk figured that the EU swung enough weight that he could afford a mild disposition. Günberk might be the real problem today, but so far he was just listening.

Okay. Alfred spread his hands. "I truly think the course we set in Barcelona is the most prudent one. Can you deny the progress we have made?" He waved at the biographical reports scattered around the table. "We have hands and minds on the scene — all deniable, and ignorant of what is manipulating them. In fact, they totally misunderstand the significance of this operation. Do you doubt this? Do you think that the Americans have any whiff of our investigation?"

Both youngsters shook their heads. Keiko even gave him a rueful smile. "No. Your SHE-based compartmentalization is truly a revolution in military affairs."

"Indeed, and our releasing those methods — even to sister services within the Alliance — shows how seriously we at the EIA view the current necessities. So, please. If we delay more than one hundred hours, we might as well start over. What is your problem with giving the final go-ahead?"

Günberk glanced at his Japanese counterpart. She made an impatient gesture for him to go ahead. "I assume your question is rhetorical, Alfred. The problem with Plan Rabbit is Rabbit. Everything depends on him, and still we know almost nothing about him."

"And neither will the Americans. Deniability is the whole point. Rabbit is everything we could want."

"He is more, Alfred." Günberk's gaze was steady. For all his youth, Braun had the stolid aspect of a turn-of-the-century German. He moved from point to point slowly, inexorably. "In setting up this operation, Rabbit has performed miracles on our behalf. His ability demonstrates that he himself is a threat."

Vaz glanced at the results of Günberk's latest investigation. "But you have discovered critical weaknesses in Rabbit. However much he's tried to disguise it, you've traced all his certificate authority to a single apex." Having a single CA apex was not unusual; that Günberk had managed to discover Rabbit's apex was a triumph. For Alfred — given his own, ah, sensitive relationship with Rabbit — it was miraculously good news.

Günberk nodded. "Credit Suisse. So what?"

"So if Rabbit turns out to be a nightmare, you could pull the plug on Credit Suisse and put him out of business."

"Pull the plug on Credit Suisse CA? Do you have any idea what that would do to the European economy? I'm proud of my people, that they ferreted this secret out — but it's not something we can effectively use."

"We should have dropped Rabbit after that first meeting in Barcelona," said Keiko. "He is too clever."

Vaz raised a hand, "Perhaps, but how could we know?"

" Ja ? Forgive me, Alfred, but we wonder if you know more about Mr. Rabbit than we."

Damn ! "Not at all. Honestly." Alfred leaned back in his chair and took in the nervous postures of his colleagues. "You've been talking behind my back, haven't you?" He gave them a gentle smile. "Do you think Rabbit is really American intelligence? Chinese?" They had spent a lot of time investigating those possibilities. But now Keiko shook her head. "Then what is your theory, my friends?"

"Well," said Günberk, sounding a little embarrassed. "Maybe Mr. Rabbit is not even human. Maybe it's an Artificial Intelligence."

Vaz laughed. He glanced at Keiko Mitsuri. "And you?"

"I think AI is a possibility we should consider. Rabbit's talents are so broad, his work is so effective — and his personality is so juvenile. That last was one of the features the U.S. DARPA thought would be characteristic." She saw the incredulity on Vaz's face. "Not every threat is a cult or conspiracy."

"Of course. But AI monsters? That's a bogeyman out of the twentieth century. Who in the intelligence communities takes that seriously? Ah! That's Pascal Heriot's hobbyhorse, isn't it?" Alfred's tone became low and serious. "Have you been talking to Pascal about this project?"

"Of course not. But AI is a threat that's been totally overlooked in recent years."

"Correct, because nothing ever came of it. Before the Sino-American war, we know DARPA spent billions on the Little Helper Project. It was almost as much a fiasco as their Space Access Denial initiative."

"Space Denial worked ."

Vaz laughed. "It worked against everybody, Keiko, the Americans most of all. But you're right, SAD is not a proper comparison. My point is that some of the smartest people in the world tried to create AI and failed."

"The researchers failed, but surely runnable code survived. The Internet is not the cramped toy it once was. Maybe pieces of DARPA's Little Helper are out there, growing into what it could never be in the low-tech past."

"That is science fiction! There was even a movie — "

"More than one, actually," said Günberk. "Alfred, I don't agree with Keiko that programs from years ago could self-organize just because decent resources are available now. But here at the IB, we have been tracking the possibilities. I think Pascal Heriot has a point. Just because most people have dismissed the possibility doesn't mean that it is not real. We are certainly past the crossover point when it comes to computer hardware. Pascal thinks that when it finally happens, it will arise without institutional precursors. It will be like many research developments, but rather more catastrophic." Just another way humankind might fail to survive the century.

"Whatever the explanation," said Keiko, "Rabbit is simply too competent, too anonymous… I'm sorry, Alfred, we think the operation should be shut down. Let's approach our American friends on this."

"But equipment is in place. Our people are in place."

She shrugged. "With Rabbit managing things? That could leave Rabbit with whatever we discover in San Diego. Even if we agreed with you, our bosses would never go along."

She was serious. Alfred glanced at Braun. He was, too. This was bad. "Keiko, Günberk, please. Just balance the risks."

"We are," said Keiko. "Rabbit loose within this grandiose scheme is a cosmic-sized un safety!" She could be quite full of modern Japanese bluntness.